On Mon, 2017-03-13 at 12:45 +0100, Diego Biurrun wrote: > On Sat, Mar 11, 2017 at 11:33:31AM +0100, Luca Barbato wrote: > > On 09/06/2016 17:12, Diego Biurrun wrote: > > > From: Alexandra Hájková <[email protected]> > > > > > > --- > > > libavcodec/vorbis_parser.c | 32 +++--- > > > libavcodec/vorbisdec.c | 237 +++++++++++++++++++++++------ > > > ---------------- > > > 2 files changed, 135 insertions(+), 134 deletions(-) > > > > Uoti pointed out that might be nicer to have it in sooner than > > later. > > This needs further performance investigation first. All x86 > benchmarks > show a slowdown, in both 32 and 64 bits.
I don't think the slowdown matters that much. Even on 32-bit where the change is more noticeable (but still below 10%), it seems that the codec is still fast enough that a benchmark uses >10% of time in things like memory copying and sample format conversion. In other words, unless you're going to do low-level optimization, it should be "fast enough". And since Vorbis performance really isn't much of a practical issue for anyone, it's not likely that waiting will result in anyone investigating the speed change or trying to improve it. So unless someone intends to optimize it, or has an argument why minor vorbis speed loss would really matter in practice and would be worth spending effort on, I think the practical thing to do would be to just apply the change. _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
